THE REMAINING 40% OF RAMANUJAN’S LOST NOTEBOOK

Bruce C. Berndt

To provide the setting for the material discussed in the sequel, we briefly describe
the history of Ramanujan’s lost notebook. It will be convenient to begin in the year
1976 and then proceed backward.
In the spring of 1976, George Andrews visited Trinity College, Cambridge, to
examine the papers left by G. N. Watson. Among Watson’s papers, he found a
manuscript containing 138 pages in the handwriting of Ramanujan. In view of
the fame of Ramanujan’s notebooks, published in a photocopy edition by the Tata
Institute of Fundamental Research in Bombay in 1957 [14], it was natural to call this
newly found manuscript “Ramanujan’s lost notebook.” How did this manuscript
reach Trinity College?
Watson died in 1965 at the age of 79. Shortly thereafter, on separate occasions,
J. M. Whittaker and R. A. Rankin visited Mrs. Watson. The late J. M. Whit-
taker was a physicist and son of E. T. Whittaker, who coauthored with Watson
probably the most popular and frequently used text on analysis in the 20th cen-
tury [25]. Rankin, who has been at the University of Glasgow for many years, had
succeeded Watson as Professor of Mathematics at the University of Birmingham,
where Watson served for most of his career. Both Whittaker and Rankin went to
Watson’s attic office to examine the papers left by him, and both found the afore-
mentioned manuscript by Ramanujan. Rankin suggested to Mrs. Watson that her
late husband’s papers be sorted and sent to Trinity College Library, Cambridge, for
preservation. During the next three years, Rankin sent Watson’s papers, including
the Ramanujan manuscript, sent on 26 December 1968, in batches to Trinity’s li-
brary. Rankin had not realized the importance of Ramanujan’s manuscript and so
did not mention it to anyone. In particular, he did not reveal the manuscript’s ex-
istence in his obituary [17] of Watson written for the London Mathematical Society.
Thus, the next question is: How did Watson come into possession of this sheaf of
138 pages of Ramanujan’s work?
After Ramanujan died in 1920, G. H. Hardy strongly urged that Ramanujan’s
published papers, notebooks, and other unpublished work be collected together for
publication. A handwritten copy of Ramanujan’s notebooks [14] was shipped from
the University of Madras to Hardy in 1923, and at the same time other manuscripts
and papers of Ramanujan were also sent. There apparently is no record of precisely
what was included in this shipment. Thus, most likely, the lost notebook was sent
to Hardy in 1923.
In the 1920s and 1930s, Watson wrote almost 40 papers on the work of Ra-
manujan, most of them arising from either Ramanujan’s letters to Hardy or from
Ramanujan’s notebooks. In particular, he wrote on Ramanujan’s mock theta func-
tions, which Ramanujan discovered in the last year of his life and described in a
Typeset by AMS-TEX
1
2 BRUCE C. BERNDT

letter to Hardy only about three months before he died [7, pp. 220–223]. Wat-
son had made some conjectures on the existence of certain mock theta functions.
If he had had in his possession the lost notebook, he would have seen that his
conjectures were correct. Thus, probably sometime after Watson’s interest in Ra-
manujan’s work waned in the late 1930s, but before his death in 1947, Hardy passed
Ramanujan’s manuscript to Watson. Since many identities involving mock theta
functions appear in the lost notebook, it is certain that the lost notebook arises
from the last year of Ramanujan’s life.
Finally, in early 1988, just after the centenary of Ramanujan’s birth, Narosa
Publishing House in New Delhi published a photocopy edition of the lost notebook
[16]. Included in the edition are a few other unpublished manuscripts by Ramanujan
as well as letters between Ramanujan and Hardy. Shortly thereafter, Rankin wrote
a very interesting paper on the origin and content of the lost notebook, as well as
other manuscripts left by Ramanujan [18].
About 60% of the approximately 650 claims made by Ramanujan in his lost
notebook pertain to mock theta functions and other q–series. Most of these results
have now been proved by Andrews. We cite just one of his papers [1], which
provides some of the lost notebook’s history that we have related above. The
remaining 40% is devoted mostly to topics examined by Ramanujan in the ordinary
notebooks. For example, theta function identities, modular equations, Eisenstein
series, integrals of theta functions, incomplete elliptic integrals of the first kind, the
Rogers–Ramanujan continued fraction, and other continued fractions are some of
the topics found in the lost notebook. This 40% (as well as the other 60%) is of great
interest to the present author. In the remainder of this paper, we briefly describe
some of these results. Much of the research on these entries is being conducted with
the author’s recent and current graduate students, in particular, Heng Huat Chan,
Sen–Shan Huang, Soon–Yi Kang, Wen–Chin Liaw, Jaebum Sohn, Seung Hwan Son,
and Liang–Cheng Zhang.
Definitions
As customary, set
∞
Y
(a; q)∞ := (1 − aq n ), |q| < 1.
n=0

Ramanujan’s general theta function f (a, b), which has the same generality as the
general classical theta function, as found in [25, Chap. 21], for example, is defined
by
∞
X
(1) f (a, b) := an(n+1)/2 bn(n−1)/2 , |ab| < 1.
n=−∞

In Ramanujan’s notation, the three most important special cases of (1) are
∞
X 2
ϕ(q) :=f (q, q) = qn ,
n=−∞
X∞
1
ψ(q) := f (q, 1) = q n(n+1)/2 ,
2 n=0
 THE REMAINING 40% OF RAMANUJAN’S LOST NOTEBOOK 3

and
∞
X
2
(2) f (−q) :=f (−q, −q ) = (−1)n q n(3n−1)/2 = (q; q)∞ =: q −1/24 η(τ ),
n=−∞

where q = exp(2πiτ ), Im τ > 0, and η denotes the Dedekind eta–function. The

penultimate equality
√ in (2) is Euler’s pentagonal number theorem.
If q = exp(−π n), for some positive rational number n, The Ramanujan–Weber
class invariant Gn is defined by

(3) Gn := 2−1/4 q −1/24 (−q; q 2 )∞ .

The Rogers–Ramanujan continued fraction R(q) is defined by

1
q5 q q2 q3
(4) R(q) := , |q| < 1.
1 + 1 + 1 + 1 + ···

Theta Function Identities

In his lost notebook, Ramanujan refined or factored some of the theta function
identities recorded by him in his notebooks. For example, in his notebooks [2, p.
262, Entry 10], Ramanujan recorded the identity

(−q 5 ; q 10 )∞
ϕ2 (q) − 5ϕ2 (q 5 ) = −4f 2 (−q 2 ) .
(−q; q 2 )∞

On page 56 in √
his lost notebook,
√
he factored this identity into the following identi-
1− 5 1+ 5
ties: If α = 2 and β = 2 , then
√
√ 5 (1 + 5)f (−q 2 )
ϕ(q) + 5ϕ(q ) = Y ¡ ¢ Y ¡ ¢,
1 + αq n + q 2n 1 − βq n + q 2n
nodd neven
√
√ (1 − 5)f (−q 2 )
ϕ(q) − 5ϕ(q 5 ) = Y ¡ ¢ Y ¡ ¢.
1 − αq n + q 2n 1 + βq n + q 2n
neven nodd

In his notebooks [3, pp. 12, 13, Entry 1], Ramanujan introduced the parameter

k := R(q)R2 (q 2 ),

and stated two elegant modular equations,

µ ¶2 µ ¶
5 1−k 5 2 2 1+k
R (q) = k and R (q ) = k .
1+k 1−k

In his lost notebook, Ramanujan gives several other beautiful identities involving
the parameter k, for example,
µ ¶5
k 1 + k − k2
= q(−q; q)24
∞.
1 − k2 1 − 4k − k 2
4 BRUCE C. BERNDT

Thus, Ramanujan has given a beautiful paramtrization for the discrimiant function,
∆(τ ) := q(q : q)24
∞ , where q = exp(2πiτ ).
Proofs of the results cited above have been given by Kang [11].
On page 206 in his lost notebook, Ramanujan gave a very strange sequence of
septic identities, which we now quote. Let

ϕ(q 1/7 )
= 1 + u + v + w.
ϕ(q 7 )

Then

8q 2 (−q; q 2 )∞
(5) p = uvw =
(−q 7 ; q 14 )7∞

and
ϕ8 (q) ϕ4 (q)
− (2 + 5p) + (1 − p)3 = 0.
ϕ8 (q 7 ) ϕ4 (q 7 )
Furthermore,

µ ¶1/7 µ ¶1/7 µ ¶1/7

α2 p β2p γ2p
u= , v= , w= ,
β γ α

where α, β, and γ are roots of the equation

µ ¶
3 2ϕ4 (q)
x + 2x 1 + 3p − 4 7 + xp2 (p + 4) − p4 = 0.
ϕ (q )

For example,
√ √ n o
(6) ϕ(e−7π 7
) = 73/4 ϕ(e−π 7
) 1 + (−)2/7 + (−)2/7 + (−)2/7 .

Although u, v, and w are not given by Ramanujan, they are given in Entry 17(iii)
in Chapter 19 of Ramanujan’s second notebook [2, p. 303]. Explicitly,

f (q 5 , q 9 ) f (q 3 , q 11 ) f (q, q 13 )
u = 2q 1/7 , v = 2q 4/7 , and w = 2q 9/7 .
ϕ(q 7 ) ϕ(q 7 ) ϕ(q 7 )

Equality (5) is easy to prove, but the proofs of the remaining identities are very
difficult. Except for (6), Son [22] has found proofs of the remaining claims. Did
Ramanujan have explicit identifications for the three missing expressions√in (6), or
did he merely conjecture that there exists a representation for ϕ(e−7π 7 ) in the
form (6)? It seems to us that if Ramanujan were able to complete this entry, he
would have needed to know the value of the class invariant (see (3)) G343 , which
apparently he did not know, and which we also do not know.
The papers by Kang [10], [11] and Son [20]–[23] contain proofs of several further
theta function identities from the lost notebook.
 THE REMAINING 40% OF RAMANUJAN’S LOST NOTEBOOK 5

Lambert Series
Pages 353–357 in the publication of the lost notebook [16] comprise a summary
of some of Ramanujan’s findings about Lambert series. The paper that we will
write for the published proceedings of the RIMS Symposium on Number Theory
will concentrate on this fragment. Here we cite only two examples.
Beginning with the work of Gauss, Legendre, and Jacobi, Lambert series have
been useful in deriving formulas for the number of representations of a positive
integer n by certain quadratic forms. For example, if rk (n) denotes the number of
ways of representing the positive integer n as a sum of k squares, then the Lambert
series identity
X∞
8 n3 q n
ϕ (q) = 1 + 16 ,
n=1
1 − (−q)n
due to Jacobi, yields the formula
X
r8 (n) = 16(−1)n (−1)d d3 .
d|n

Similarly, let tk (n) denote the number of ways that the positive integer n can be
represented as a sum of k triangular numbers. Then the Lambert series identity
∞ ∞
3/2 6 2 1 X (2n + 1)2 q (2n+1)/2 1 X (−1)n (2n + 1)2 q (2n+1)/2
q ψ (q ) = −
16 n=0 1 + q 2n+1 16 n=0 1 − q 2n+1

gives the corollary

1 X 1 X
t6 (n) = d2 − d2 .
8 8
d|(4n+3) d|(4n+3)
d≡3 (mod 4) d≡1 (mod 4)

We have been unable to find this formula in the classical literature. The first
occurrences known to us are in recent papers by Kac and Wakimoto [9] and by
Ono, Robins, and Wahl [12]. We have found a proof along the lines of Ramanujan’s
thinking.
Incomplete Elliptic Integrals of the First Kind
Some of the most amazing formulas in the lost notebook involve incomplete
elliptic integrals. We cite just one of several examples. Recall that f (−q) is defined
by (2).
If
f 3 (−q)f 3 (−q 15 )
v := q 3 ,
f (−q 3 )f 3 (−q 5 )
then
Z Z √
q 2 arctan(1/ 5)
1 dt
(7) f (−t)f (−t3 )f (−t5 )f (−t15 )dt = ³ q ´q .
0 5 2 arctan 1
√
1−11v−v 2
1− 9
sin2 t
5 1+v−v 2 25

This and several other formulas in the lost notebook of the same sort were first
proved by Raghavan and Rangachari [13]. However, their proofs partially depend
6 BRUCE C. BERNDT

upon ideas with which Ramanujan would not have been familiar. In particular, to
prove (7), they used the remarkable differential equation
dv p
(8) = f (−q)f (−q 3 )f (−q 5 )f (−q 15 ) 1 − 10v − 13v 2 + 10v 3 + v 4 ,
dq
which they quote from the treatise of Fricke [8, p. 439]. The author, H. H. Chan,
and S.–S. Huang in an unpublished manuscript have given more elementary proofs
of (7) and (8).
The Rogers–Ramanujan Continued Fraction
Recall that the Rogers–Ramanujan continued fraction R(q) is defined by (4).
Using the Rogers–Ramanujan identities, we obtain the beautiful representation
(q; q 5 )∞ (q 4 ; q 5 )∞
(9) q −1/5 R(q) = .
(q 2 ; q 5 )∞ (q 3 ; q 5 )∞
But perhaps the most important and useful theorem about R(q) is given by the
formula
1 f (−q 1/5 )
(10) − 1 − R(q) = 1/5 ,
R(q) q f (−q 5 )
where f (−q) is defined by (2). The corollary,
1 5 f 6 (−q)
(11) − 11 − R (q) = ,
R5 (q) qf 6 (−q 5 )
is also very useful. These results are found in Ramanujan’s second notebook [14,
pp. 265–267] and were first proved by Watson [24] for the purpose of establishing
some of Ramanujan’s claims about R(q) made in his first two letters to Hardy [15,
pp. xxvii, xxviii]. They were also crucially used by the author, Chan, and Zhang
[5] in deriving general formulas for the explicit evaluation of R(q). As we shall see
in the next paragraphs, Ramanujan recorded two–variable generalizations of (10)
and (11) in his lost notebook.
On page 207 in his lost notebook, Ramanujan listed three identities,
f (−q 1/5 , −λq 2/5 )
(12) P −Q=1+ ,
q 1/5 f (−λ10 q 5 , −λ15 q 10 )

f (−λ, −λ4 q 3 )f (−λ2 q, −λ3 q 2 )

(13) PQ = 1 − ,
f 2 (−λ10 q 5 , −λ15 q 10 )

f (−q, −λ5 q 2 )f 5 (−λ2 q, −λ3 q 2 )

(14) P 5 − Q5 = 1 + 5P Q + 5P 2 Q2 + ,
q f 6 (−λ10 q 5 , −λ15 q 10 )
without specifying the functions P and Q. Son [19] has been able to discern the
identities of P, Q, and R, and so prove the following theorem. If
f (−λ10 q 7 , −λ15 q 8 ) + λqf (−λ5 q 2 , −λ20 q 13 )
(15) P = ,
q 1/5 f (−λ10 q 5 , −λ15 q 10 )
λf (−λ5 q 4 , −λ20 q 11 ) − λ3 qf (−q, −λ25 q 14 )
(16) Q= ,
q −1/5 f (−λ10 q 5 , −λ15 q 10 )
 THE REMAINING 40% OF RAMANUJAN’S LOST NOTEBOOK 7

then (12)–(14) hold.

By setting λ = 1, in (15) and (16), using the quintuple product identity

f (−x2 , −λx)f (−λx3 )

f (−λ2 x3 , −λx6 ) + xf (−λ, −λ2 x9 ) = ,
f (−x, −λx2 )

with (x, λ) = (q, q 2 ) and (q 2 , q −1 ), respectively, employing Jacobi’s triple product

identity, and utilizing (9), we see that

f (−q 7 , −q 8 ) + qf (−q 2 , −q 13 ) f (−q 2 , −q 3 ) 1

(17) P = 1/5 5
= 1/5 4
= ,
q f (−q ) q f (−q, −q ) R(q)
f (−q 4 , −q 11 ) − qf (−q, −q 14 ) q 1/5 f (−q, −q 4 )
(18) Q= = = R(q).
q −1/5 f (−q 5 ) f (−q 2 , −q 3 )

Since P Q = 1, from (17), (18), and (2), we see that (12) and (14) reduce to the
two main identities, (10) and (11), respectively.
We close with a beautiful transformation for a generalization of the Rogers–
Ramanujan continued fraction
√ found on page 46 of the
√ lost notebook.
Let k ≥ 0, α = (1 + 1 + 4k)/2, and β = (−1 + 1 + 4k)/2. Then, for |q| < 1
and Re q > 0,
(17)
1 k+q k + q2 k + q3 1 q q2 q3
= .
1+ 1 + 1 + 1 + ··· α + α + βq + α + βq 2 + α + βq 3 + · · ·

In particular, if k = 2, we obtain the following elegant corollary, also found on page

46, but with a slight misprint. For |q| < 1,

1 2+q 2 + q2 2 + q3 1 q q2 q3
= .
1 + 1 + 1 + 1 + ··· 2 + 2 + q + 2 + q2 + 2 + q3 + · · ·

For proofs of many results on the Rogers–Ramanujan continued fraction found

in the lost notebook, see a paper by the author, Huang, Sohn, and Son [6], and for
a survey on the Rogers–Ramanujan continued fraction, with emphasis on results in
the lost notebook, see [4].

References
1. G. E. Andrews, An introduction to Ramanujan’s “lost” notebook, Amer. Math. Monthly 86
(1979), 89–108.
2. B. C. Berndt, Ramanujan’s Notebooks, Part III, Springer–Verlag, New York, 1991.
3. B. C. Berndt, Ramanujan’s Notebooks, Part V, Springer–Verlag, New York, 1997.
4. B. C. Berndt, H. H. Chan, S.–S. Huang, S.–Y. Kang, J. Sohn, and S. H. Son, The Rogers–
Ramanujan continued fraction (to appear).
5. B. C. Berndt, H. H. Chan, and L.–C. Zhang, Explicit evaluations of the Rogers–Ramanujan
continued fraction, J. Reine Angew. Math. 480 (1996), 141–159.
6. B. C. Berndt, S.–S. Huang, J. Sohn, and S. H. Son, Some theorems on the Rogers–Ramanujan
continued fraction in Ramanujan’s lost notebook (to appear).
7. B. C. Berndt and R. A. Rankin, Ramanujan: Letters and Commentary, Amer. Math. Soc.,
Providence, 1995; London Math. Soc., London, 1995.
8. R. Fricke, Die Elliptische Funktionen und ihre Anwendungen, Bd. II, Teubner, Leipzig, 1922.
9. V. G. Kac and M. Wakimoto, Integrable highest weight modules over affine superalgebras and
number theory, Lie Theory and Geometry (J.–L. Brylinski, R. Brylinski, V. Guillemin, and
V. Kac, eds.), Birkhäuser, Boston, 1994.
8 BRUCE C. BERNDT

10. S.–Y. Kang, Some theorems on the Rogers–Ramanujan continued fraction and associated
theta function identities in Ramanujan’s lost notebook (to appear).
11. S.–Y. Kang, Ramanujan’s formulas for the explicit evaluation of the Rogers–Ramanujan con-
tinued fraction and theta–functions (to appear).
12. K. Ono, S. Robins, and P. T. Wahl, On the representation of integers as sums of triangular
numbers, Aequa. Math. 50 (1995), 73–94.
13. S. Raghavan and S. S. Rangachari, On Ramanujan’s elliptic integrals and modular identities,
Number Theory and Related Topics, Oxford University Press, Bombay, 1989, pp. 119–149.
14. S. Ramanujan, Notebooks (2 volumes), Tata Institute of Fundamental Research, Bombay,
1957.
15. S. Ramanujan, Collected Papers, Chelsea, New York, 1962.
16. S. Ramanujan, The Lost Notebook and Other Unpublished Papers, Narosa, New Delhi, 1988.
17. R. A. Rankin, George Neville Watson, J. London Math. Soc. 41 (1966), 551–565.
18. R. A. Rankin, Ramanujan’s manuscripts and notebooks, II, Bull. London Math. Soc. 21
(1989), 351–365.
19. S. H. Son, Some theta function identities related to the Rogers–Ramanujan continued fraction,
Proc. Amer. Math. Soc. (to appear).
20. S. H. Son, Cubic identities of theta functions, The Ramanujan J. (to appear).
21. S. H. Son, Some integrals of theta functions in Ramanujan’s lost notebook, Proceedings of the
Fifth Canadian Number Theory Association Meeting (R. Gupta and K. S. Williams, eds.),
American Mathematical Society, Providence, RI (to appear).
22. S. H. Son, Septic theta function identities in Ramanujan’s lost notebook (to appear).
23. S. H. Son, Two theta function identities in Ramanujan’s lost notebook (to appear).
24. G. N. Watson, Theorems stated by Ramanujan (VII): Theorems on continued fractions, J.
London Math. Soc. 4 (1929), 39–48.
25. E. T. Whittaker and G. N. Watson, A Treatise on the Theory of Bessel Functions, 2nd ed.,
Cambridge University Press, Cambridge, 1966..

Department of Mathematics, University of Illinois, 1409 West Green Street,

Urbana, Illinois 61801, USA
E-mail address: berndt@math.uiuc.edu